Incomplete block design

I have data among several variables sampled from deep-sea sites located in a marine canyon. In total 13 sites (n=3/site) were sampled. Sampling was conducted on 5 sites along 1 transect (~ 500 - 2200 m), 5 sites on another transect (~500 - 1800 m), and 3 sites on another transect (~ 770 - 1000 m). Each transect represents a different treatment. The sampling design is unbalanced and it has been suggested that incomplete block design could be used to test for differences among my treatments. Is this a valid suggestion? What would my blocking factor be? New to blocking designs so I would appreciate any insight. Thanks in advance.


Active Member
I don't really see it either. The blocking factor would have to be something that had multiple treatments on it, at a minimum. So not transect. Could it be site? I would think that site is nested in transect, that is that the effect of site 1 is not really expected to be the same on each transect?


TS Contributor
This appears to be an unbalanced split-plot design where transect is the whole plot and sites are the subplots. There are several good videos from SAS JMP that explain the concept. Don't get turned off by the manufacturing examples. Split plot designs originated in agriculture.